HarnessmarketEnter the desk
/HMX Claude Code 29.1%/Open index 519 slips/Meridian desk 6,400 agents live/Skills + MCP 553,855 indexed (Skillful, Aug 2026)/Agent Skills spec 1.2M+ open packages/Deloitte: orchestration worth +15–30% of autonomous-agent TAM by 2030/Gartner: 40% of agentic projects cancelled by 2027 without a supervisor/SWE-bench Pro: harness swap > many model upgrades/HM-SWE-Pro: Loop+Claude 41.2 pass@1 · chat 18.2 · same model/Lewis 2608.26218: F2PF 28→49 under a tighter harness, same model/ACES: Skill Lift 0.21 · 947 paired cases · scan vs live ρ=0.14/Tokenomics: code review 59.4% of ChatDev tokens/HMX Claude Code 29.1%/Open index 519 slips/Meridian desk 6,400 agents live/Skills + MCP 553,855 indexed (Skillful, Aug 2026)/Agent Skills spec 1.2M+ open packages/Deloitte: orchestration worth +15–30% of autonomous-agent TAM by 2030/Gartner: 40% of agentic projects cancelled by 2027 without a supervisor/SWE-bench Pro: harness swap > many model upgrades/HM-SWE-Pro: Loop+Claude 41.2 pass@1 · chat 18.2 · same model/Lewis 2608.26218: F2PF 28→49 under a tighter harness, same model/ACES: Skill Lift 0.21 · 947 paired cases · scan vs live ρ=0.14/Tokenomics: code review 59.4% of ChatDev tokens

Skill · certified

SWE Eval Pack

Harness-aware evals. The pack that made the SWE-bench Pro finding operational.

OriginSubscribe on desk

Publisher

Lattice Labs

License

commercial-usage

Latest

1.9.0

Installs

3,100

Thesis

Eval suites that hold the model constant and vary the harness — the experiment most teams swear they ran and did not.

If you cannot produce a table where the harness is the independent variable, you are not doing harness engineering. This pack is that table, weekly.

Direction

This listing is a reference. The tree lives at origin. The command is how you load it here.

Subscribe (usage) → hm install lattice/eval-pack-swe@1.9.0

Hold the model. Change the harness. Publish the table. Anything else is anecdote.

Versions

SemverSHAReleasedNotesSigned
1.9.0swe190a2026-08-06Pro subset refreshed. Cost traces.signed · sbom

Subscribe

$0.40

Usage · eval run

Enterprise from $14,000

Subscribe, then pin. Entitlement is checked at deploy, not in a PDF.

Origin

registry.harness.market/lattice/eval-pack-swe

via registry

Open origin →

Compatibility

  • claude-code
  • codex-cli
  • grok-build
  • opencode
  • cursor-agent

Permission manifest

fs.read
**
fs.write
**
network
registry.harness.market, api.github.com
exec / secrets
exec · GITHUB_TOKEN

Publisher

Lattice Labs

Independent harness engineering. Recovery loops, schedulers, evals.